Alpha's DNA Through the Lens of Alternative Factors

If there is one thing that 2020 has taught us, it is to expect the unexpected. The tragic events that unfolded at the Capitol building this past Wednesday were no exception, and our hearts go out to the families who lost their loved ones. We can only hope that the path ahead is now a little more stable.

In last week’s Factor Spotlight, we discussed the concept of an alpha-driven stock and looked to find insights based on the median factor exposures of these stocks compared to the broad Russell 3000 universe. We confirmed an obvious conclusion that alpha opportunity deteriorated during the crisis-driven 2020 period and also highlighted key exposures of this group of stocks for traditional style factors.

This week, we’ll extend the analysis a step further by overlaying less traditional, but more unique factors on our analysis to further understand the characteristics of alpha-driven securities.

A Stricter Definition of an Alpha-Driven Stock

We’ll start by tightening our definition of an alpha-driven stock. We define an alpha-driven security as one where 67% or more of its expected risk is idiosyncratic, as defined by a risk model. Last week, we used the Axioma US 4 Medium Horizon (“Axioma US4”) risk model to measure the idiosyncratic risk. However, since one of the major benefits of the Omega Point platform is that we can easily view risk through the lens of multiple risk models and datasets so we’ll take advantage of that this week and leverage additional models in our analysis.

In addition to the Axioma US4 risk model, we’ll also look at idiosyncratic risk measured by the MSCI Barra Global Total Market Equity Model for Long-Term Investors (“Barra GEMLT”) risk model. We chose to use this model to complement the Axioma model because it is one of the more common Barra models used by institutional investors and allocators. For our purposes, we’ll define a “true” alpha-driven stock as one where both risk models determine that 67% or more of the predicted risk is idiosyncratic. We also include factor-driven securities (67% or more of expected risk is systematic, as determined by both Axioma US4 and Barra GEMLT) and neutral securities (neither factor-driven or alpha-driven, as determined by both Axioma US4 and Barra GEMLT) in the analysis.

This framework was applied to segment the Russell 3000 into the three groups (alpha-driven, factor-driven, and neutral) on a month-end basis from Dec 2019 to Dec 2020 using both aforementioned risk models. We found that the risk models classified securities the same (“agreed”) 82% of the time, on average.

The cases where the risk models agree form our “true” alpha-driven, factor-driven, and neutral stock groups and will be the universe that we will use for the remainder of the analysis.

Even when controlling for this stricter definition, we saw the same high level trends in the percentage of Russell 3000 securities falling into each of our groups. For additional discussion on these trends, please see last week’s Factor Spotlight post.

The More (Risk Models), the Merrier

Once we had our true stock groups defined, we used additional risk models to take a different view on the characteristics of alpha-driven securities. The two supplementary risk models used were the MSCI Barra US Total Market Equity Model for Medium-Term Investors (“Barra USMED”) and the Wolfe Research QES US Broad (“Wolfe QES US”) risk model. We chose these models because they incorporate unique factors that can help identify crowding trends and sensitivity to macro-economic indicators.

‍Crowding Factors

‍To measure the level of crowdedness of the securities within our stock groups, we will use the following factor exposures.

Macro Factors

‍To measure the level of sensitivity to macro-economic indicators within our stock groups, we will use the following factor exposures.

Are Alpha-Driven Names Crowded?

We overlaid the above factor exposures onto our stock groups and calculated the median exposure for each factor across our groups as well as for the total Russell 3000 universe.

‍Liquidity Factor Exposure

When using a standard definition of liquidity (recent average daily trading volume, relative to market cap), we saw last week that the alpha-driven stocks had significantly higher median Liquidity factor exposure throughout 2020 as compared to the other groups and the Russell 3000 overall. However, using the Barra USMED risk model’s alternative definition for liquidity which controls for market-wide liquidity levels, we see a different trend.

The alpha-driven stocks have a lower median Liquidity factor exposure for most of 2020 relative to the other groups, with a sharp decline during the COVID-19 market downturn in March 2020. This indicates that in fact, relative to the full level of liquidity available in the market, the alpha-driven stocks were actually fairly illiquid. This story started to change as the market rebounded and the alpha-driven names increased in relative liquidity. At the same time, the factor-driven securities, along with the neutral and overall Russell 3000, steadily declined in median Liquidity factor exposure over the year. Overall, we can see that alpha-driven names had increased relative trading activity and ended the year higher than the factor-driven group and at the same level as the total Russell 3000. This is a possible indication that alpha-driven names became more crowded throughout the year.

‍Short Interest Factor Exposure

While the median Barra USMED Short Interest factor exposure was low and relatively unchanged for the factor-driven, neutral, and total Russell 3000 groups, there was a considerable trend change for the alpha-driven group. It appears that alpha-driven names were not being heavily shorted pre-COVID-19; however, this reversed course in a big way, as exemplified by the large jump in the median exposure for this group from -0.19 in Feb 2020 to 0.36 in March 2020. The median exposure continued to climb for alpha-driven names until May 2020, when the level of short interest started to re-normalize. The alpha-driven group finished 2020 with a relatively high median Short Interest factor exposure, indicating that short sellers are continuing to be very bearish on the downside performance potential for these names.

‍HF Crowding Factor Exposure

Looking at the long side of crowdedness, the median Wolfe QES US HF Crowding exposure shows an interesting trend. The median exposure for the factor-driven group was on a steady decline, indicating that these names became less sensitive to hedge fund concentration. The alpha-driven group showed the opposite trend, with the median factor exposure on the rise and peaking in July 2020. After that, the median exposure plummets, with the alpha-driven group ending 2020 with the lowest sensitivity to hedge fund concentration of all of the groups. This points to alpha-driven names becoming less sensitive to crowding on the long side by hedge fund investors.

Are Alpha-Driven Names Sensitive to Macro Indicators?

Oil Beta Factor Exposure

The beginning of 2020 saw the alpha-driven stocks with a slightly positive median exposure to the Wolfe QES US Oil Beta factor and the factor-driven stocks with a slightly negative median exposure. However, these trends fluctuated heavily over the year. The alpha-driven names saw heavy variation in exposures, hitting a peak median exposure of 0.53 in March 2020, then plummeting back to pre-COVID levels in April 2020, just before rising again and maintaining elevated levels until Nov 2020.

The factor-driven names saw the median factor exposure turn positive during the March market turmoil, but this trend quickly reversed. Despite a small jump to positive median exposure in Nov 2020, the median Oil Beta exposure for the factor-driven group remained negative throughout the latter half of the year.

The entire universe is on the downturn again in terms of median Oil Beta exposure, however, there is still a wide spread between the alpha-driven and factor-driven median exposures. The conclusion here is that alpha-driven names had elevated levels of sensitivity to changes in oil prices during the crisis periods of 2020.

‍Interest Rate Beta Factor Exposure

For the Wolfe QES US Interest Rate Beta factor, the alpha-driven stocks had an increasingly strong, albeit negative, median exposure throughout 2020, meaning that these securities have an inverse relationship to interest rate movement. On the flip side, the factor-driven group started the year with an extremely high median Interest Rate Beta exposure, which took a nosedive in March 2020, but has been steadily on the rise since. These median exposure trends imply that alpha-driven securities may be more tied to reopening themes which may be hurt as the economy recovers and interest rates begin to rise again, whereas factor-driven stocks may be more long-term steady-eddies that will benefit from rising rates and the economy reopening.

What Do Alpha-Driven Names Look Like in 2021?

Going into 2021, the median alpha-driven stock has relatively:

• higher liquidity (trending up)
• higher short interest (trending flat)
• lower HF crowding (trending down)
• higher oil beta (trending down)
• lower interest rate beta (trending flat)

Will the DNA of alpha-driven names in 2021 transpire in the same way as it did in 2020? This remains to be seen, but the above analysis gives investors a strong start of where to look for alpha.

US & Global Market Summary

US Market: 01/04/21 - 01/08/21

• U.S. exchanges hit record highs on Friday despite Wednesday’s deadly riots in the Capitol building and record-breaking high Covid-19 deaths. Investor sentiment was buoyed by the resolution of the Georgia Senate runoffs tipping control of the chamber to Democrats and improving prospects for additional fiscal stimulus under the new administration.
• The yield on 10-year Treasuries peaked over 1% for the first time since March, along with low interest rates for mortgages and loans pointed to stronger 2021 GDP growth
• On Friday the Bureau of Labor Statistics reported the U.S. economy lost 140,000 jobs in December, the first time that jobs have shrunk since April.
• Global manufacturing PMI data reflects a strong recovery in place, while the service sector suffers amidst rising COVID cases.
• JP Morgan, Citigroup and Wells Fargo are set to release fourth-quarter results on Jan. 15, among the first S&P 500 companies to post their results for the last period of coronavirus-stricken 2020.

Normalized Factor Returns: Axioma US Equity Risk Model (AXUS4-MH)

• Momentum has been on a tear over the past month and ends up as this week’s biggest winner once again, breaking into positive territory and far above its oversold status from early December.
• Earnings Yield continues to show positive movement for the third week in a row.
• Market Sensitivity slid into negative territory and now sits far below its Overbought label from just a few weeks ago.
• Volatility also turned negative and continues its freefall into the new year while Valueput on the brakes, at least for now.
• US Total Risk (using the Russell 3000 as proxy) increased by 6bps.

‍Normalized Factor Returns: Axioma Worldwide Equity Risk Model (AXWW4-MH)

• Momentum and Growth both continue their impressive runs of late and as they march deeper into positive territory in the new year.
• Profitability finally showed some signs of upward movement following a slow 1-month slide.
• Market Sensitivity, Value and Volatility continue to show weakness as we move further into January.
• Size moved deeper into negative territory as it rounded out the week as the biggest loser once again.
• Global Risk (using the ACWI as proxy) increased by 8bps.

Regards,
Alyx